Marketing Attribution Models: 43 Experts Explain How to Choose The Right One (and Mistakes to Avoid)
A critical part of marketing is knowing which campaigns are driving the most ROI. Only once you have this data can you effectively allocate resources to acquire new customers, and plan for future growth.
Many companies fly blind, and don't properly distribute credit to all the touchpoints along the customer journey.
The world of digital marketing has caused a major shift in the customer journey. Many of the legacy marketing attribution models are now ill equipped
to accurately measure the new customer journey that has become longer, with many different touchpoints.
According to Salesforce, it takes 6 to 8 touches to generate a viable sales lead. And 60% of the sales cycle is over before a buyer talks to your salesperson (Corporate Executive Board).
As more companies begin to run multiple cross channel campaigns, it gets trickier to measure the ROI for each one.
How much influence did a Twitter ad click have on that demo signup? How about the AdWords retargeting campaign? Is SEO contributing to product sales? Should more weighting be given to specific touchpoints? If most conversions come from SEO or PPC, should you axe social media campaigns altogether?
To find the answers to those important questions, marketers need to embrace multi-touch attribution marketing models to effectively manage, optimize and scale campaign performance.
What if you don't know which attribution is best for your campaign? The wrong model could give credit where it isn't deserved, while selling other higher performing channels.
To help clarify, we reached out to 43 CMOs, VPs and analytics experts from companies like Workzone, DocSend and Xactly to get their top tips for choosing the right attribution model, and share the common mistakes to avoid.
The answers are below.
But, before we dive into the responses here are some key takeaways...
5 Key Tips for Choosing the Right Marketing Attribution Model for Your Business
- Create a customer journey map - brainstorm touchpoints, channels and interactions at each stage of the buying cycle.
- Models should answer the executive question - If I spend nothing will revenue go down? Will reps or a team or the division miss their number?
- Pay close attention to lead quality - this could mean the rate at which leads became active users, and the rates they became paying customers.
- Campaign objectives should be a key driver - a first-touch model could work for a branding campaign, whereas a time-decay model is better suited for a limited-time promotion.
- Be prepared to pivot - always test, iterate and be willing to optimize or completely change attribution models.
These are just a few of the high-level insights.
Read on to find out how the experts decide which attribution models to use for multi-touch marketing campaigns.
How to Pick the Best Attribution Model (43 Experts Weigh in)...
Before you do anything, create a new ‘virgin’ and untouched profile view.
Leave it. It’s there for reference only!
Also, if you’re using Google Analytics you need to be aware of 3 important things:
1. The default model is Last Non-direct click. Seeing loads of direct traffic? Make sure you’re using UTMs to tag your inbound visits, particularly from social media.
2. If you’re looking at Multi-channel reports they don’t use Last Non-direct, instead they consider conversion paths. Confused already? Well…
3. If you’re running AdWords – it uses greedy attribution. That’s not a real model, that’s just it’s behaviour!
Why is it greedy? Because if an ad click is present and a conversion happens, regardless of the final traffic source of the conversion, AdWords will measure a conversion. This is wrong.
Thankfully, we can fix the GA and AdWords disagreement.
I recommend setting up your conversion tracking in Analytics, and importing your conversion data to AdWords, from Analytics.
This is both quicker, easier and irons out a few kinks in discrepancies. It also means you don’t have to maintain separate pieces of code, for both platforms.
So how do you pick a model to work with?
You need to ask questions about your data:
How many touches on average do my customers have before they convert?
How long in general is it between the first visit and the final conversion?
How many channels are involved in my conversions?
How many channels are we active on and measuring?
Your multi-channel reports - Time Lag, Conversion Paths and Path Length - can help identify trends and answer these questions.
Choosing an attribution model boils down to channels and time.
But, my favourite attribution model:
Time Decay - a more sophisticated, common sense approach to attribution. Not perfect by any stretch of the imagination, but compared to the other options – Time Decay makes sense.
The model works on the concept of exponential decay, with the most heavily credited touchpoints closest to the time of conversion. It enables you to optimise marketing (and specific campaigns) that generate momentum with time.
It respects the customer journey through multi-touch and time, and shows how channels are working together.
One thing to be aware of however, is that a long sales cycle does get slightly ‘penalised’ as Time Decay discounts the earlier research phase.
The single biggest (and probably best) tip anyone can give you about attribution is to choose a model and stick to it!
The most important point to remember:
There is no right or wrong attribution model — You must align your choice with your own unique digital strategy and data.
Whatever you decide, ensure it's for the mid-to-long run. You don't want to keep switching models, else everything is going to become skewed.
To begin with? Keep it simple.
Time Decay can be deemed an ideal starting point for many. The closer the touchpoint is to conversion, the more credit it receives (and vice-versa).
Following solid experience with this model, Position Based attribution can be an ideal next step up for a more logical credit allocation.
What to avoid?
#1: Rushing your attribution modeling decision process.
#2: Misaligning your decision with customer journeys and stakeholder expectations.
#3: Sticking with Last Click attribution when your sales cycle is lengthy.
#4: Disregarding Multi-Touch attribution because Single-Touch is 'easier'.
#5: Not using the data gathered from attribution to constantly alter channel investment.
At Workzone, we’ve implemented a version of the U-Shaped or position-based model. We weigh equally the first touch to our website and the last touch before a conversion.
First click attribution helps us understand what brings people to our brand whereas last touch attribution helps us understand which channels and assets convert best.
This helps us measure the value of our top funnel content and PPC ads, while keeping tabs on where most of the conversions occur—usually our product tour page or pricing page. Because we position our product differently to personas, it also helps with following potential use cases.
· Take into consideration the industry you are in. Each industry has varying sales cycles, so it is important to pick an attribution model that aligns with the consumer purchasing journey common in your market.
· Be sure to have a clear understanding of all touchpoints. Make sure you know all the possible channels for a customer to reach you and interact with your business. This will help you make a good decision about the attribution model to choose.
· Remain open to change. As your marketing activity evolves and your strategies are refined, it is important to be open to the idea that the attribution model you choose may also need to be updated in the future.
Common Mistakes Marketers Need to Avoid
· Not utilising a large data set to gain an accurate understanding of all the customer touchpoints with your business.
· Using a “one size fits all” approach with attribution models. Every business and every industry is different!
· Forgetting the importance of continuous monitoring and analysis. Customer journeys and interactions do change, so be sure to understand how buyer behaviour in your market is shifting.
Selecting the right attribution model is almost always a function of your reporting and insights requirements. With that, the most common mistake our clients make is not clearly identifying the attribution model which balances the following components;
What is useful - Your attribution model must be as mature as your organization. If your company is at the beginning of their digital transformation then perhaps Multi-Touch or Custom Attribution models are too advanced and too soon. Don't worry so much about what the industry is telling you, decide what type of data you need to tell your story and make decisions off of.
What is achievable - Clients will have more wins if they make specific gains in analytics and reporting, than by trying to boil the ocean. Building a culture of analytics is a hard and slow process. Think hard about which model would be the easiest to achieve in your organization knowing its culture and its goals. Make sure its a variable in your decision making.
What is scalable - The right model for you today, if you are doing things right, may or may not be the right model for you tomorrow. Make sure you do not construct your org tightly around any one model. Create an architecture that is flexible and scalable.
Attribution modelling, in my opinion, is going to be a huge differentiating factor for 2017, as companies with attribution protocols in place have a distinct advantage over those that are flying blind. That said, implementing them can seem overwhelming if you're new to the practice.
Any attribution model is better than no model. At Louder Online, we work with mostly enterprise-level clients who have huge budgets and teams to put these models into practice. But if you're a small business and multi-touch modelling is out of your reach, going with a first-touch or last-touch single attribution modelling practice is still going to give you valuable information.
Think about how you want to use your data, as this should inform the specific protocol you use (this, and the specific modelling tools you have access to). If you have a long sales cycle, for example, a time-decay multi-touch model will help you understand where and when in your funnel leads break out of nurturing stages.
If all you want to know is which of your salespeople or campaigns closed sales, last-touch, single-touch attribution will be fine.
As far as what to avoid, I'd say the biggest mistake marketers make is ignoring the data they generate, either because they forget it's there, they run out of time or they don't understand what to do with it.
Make sure your client and internal team has a clear focus on the driver strategy for the website.
How do they want to allocate the budget that is spent on each channel?
What is the target cost per lead? If there are multiple drivers and touch points you should blend the value of the lead across all the channels. If it is a short one-click Ad to purchase it should be last click.For a multitouch example:
1 - TOUCHPOINT: User sees a search Adwords ad and clicks on it.
2 - SOFT LEAD: User signs up for a newsletter or a download with their email address.
3 - RETARGETING: User later sees a targeted Facebook ad or promoted post for your core product or service based on the targeting setup from their email address you received in the Adwords Ad click for Facebook.
4 - User clicks on the Facebook Ad.
5 - CONVERSION: User signs up and pays for your product or service
The value for this customer should be spread across AdWords and Facebook. Blended attribution should be used.
However if it's a quick smaller purchase cycle:
1 - TOUCHPOINT: User sees ad.
2 - CONVERSION: User Clicks ad and purchases product or service.
Last click attribution should be used.
Bums on seats! So many times, over the last decade I have seen so many marketers continue using attribution metrics that are meaningless if they do not measure also the ultimate KPI, bums on seats, which for most industries is confirmed sales and for the education industry is usually enrolled students.
My tips for choosing the right model:
1. Understand the end to end journey.
2. Make sure every interaction is recorded and tagged properly.
3. Test, test, test, keep trying new tactics to increase volume at the top of the funnel.
4. When measuring every interaction, always differentiate your tags by media, product, month, and year as a minimum.
5. Standardize the use of tags and keep a database with all historical tags used.
6. Simplify report by having each stage and the relevant conversion rate into the next stage.
7. Realize early on the equation that top of the funnel metrics are relevant but with the revolution of digital marketing they could look high, yet be pointless down the funnel.
8. Practice AB testing along the funnel.
An attribution model is custom for each company, but getting it right will maintain your highest ROI. Some tips are to never attribute success based on prejudice.
If you attribute success to something tied to your ego, nepotism/friendships, personal investments or systems, you are almost sure to suffer and not just because you may apply funding to the wrong/assumed/hoped for top contributors but because of the potential harm it will do to your staff/teams.
I think everyone, will at some time in their life, create something wonderful or perform a hard fought battle to excel then have someone else take credit for your efforts. Does that inspire you to do it again or even put your creative juices to task in order to surpass your last effort?
When you create your attribution model, make sure it rewards the actual top performers or systems creating the greatest benefit for you company. Doing so will inspire a healthy competition getting others to strive for the top.
Otherwise, morale will suffer and so will the performance of the rest of your teams.
Choose the right marketing attribution model based on what makes sense for your company. At JotForm, we primarily look at last-touch data: where did the user sign up from? It could be organic, AdWords, social, referring websites, in-product, etc. From that source, we also look at the quality of those signups, which for us means the rates that they became active users and the rates that they became paid users. This allows us to analyze which sources are of higher quality and what is working.
It's important to recognize that, when looking at last-touch data, many factors contributed to the journey of a user's signup. Just because content marketing may not directly drive a high percentage of signups doesn't mean that it is not effective.
Many users, especially those that would sign up for a B2B software product, are doing their research prior to signing up. That includes reading reviews, reading how-to guides and other articles, and consuming comparison content. It also may be helping SEO efforts, which leads to organic search bringing in more leads.
What common mistakes do marketers need to avoid?
Marketers need to avoid the trap of making decisions based on data that doesn't support the whole picture.
For example, making a change to a landing page test that could drive a higher number of signups, and that could prematurely lead to a decision to declare that change a success and implementing it.
But were the signups of higher quality than another version? Did the change accomplish the desired result- which is not to drive more signups but better signups? It could be that one version drives less signups but those signups are more likely to upgrade to a paid plan and become active users, and the other version could drive many more signups but those new users may be more likely to remain on a free plan. More angles to the story should be considered.
For me, the right marketing attribution model has to map how you actually do business and (most importantly) close leads. As the agency began getting into marketing attribution both for ourselves and our clients, it became really easy to have too many marketing attribution channels.
The software we used gave a dozen or more attribution channels and you can easily become distracted by these things and it clutters your goal, finding the best and most profitable channels to focus on.
The next mistake is to take the opposite approach. It's easy to over-simplify the process and have it tell you things you already know. You need to have enough distinction between your attribution channels that it actually tells you some actionable information. (ie. you get leads through search, not just AdWords vs. "these are all web leads").
You also need to know which of those leads convert to sales. It's not enough to know you get a lot of leads from a source, they should be leads that grow your business. The same holds true of referral leads. Are they true "word of mouth" referrals, back link referrals, or just your reputation? This matters and having that distinction helps you know where to grow relationships.
The last tip, make sure it's automatically accounted for in your CRM. Find a CRM that can tie into your marketing software. If you leave it up to manual data entry you'll either (best case scenario) muddy up data, or worst not do any data entry at all.
There is no "right" marketing attribution model that fits every company's needs. How you focus your efforts depends your ultimate goals. And not taking the time to understand your end goals is the biggest mistake you can make when choosing an attribution model.
Some factors to take into consideration when selecting a model (or combination of models) include:
Your audience personas - Who do you want to reach? Are your audience members tech savvy? Where do they reside online? What prompts them request a demo or make a purchase?
Your sales cycle - Are you a B2B vs B2C company? Are you aiming for a click-to-buy or will the sell require a longer cycle? How important is nurturing in your cycle? Do leads need to be warmed before they purchase or request a demo?
Your vertical focus - What is important to to your audience? Where do they need to see you in order for you to be credible?
Your budget - How much can you spend on your promotion efforts? How many leads do you need in order for your campaigns to be viable? Is it about volume or targeting?
Your team - Are you a big team or a small team? How many channels can you focus on and yet not stretch yourself to a point where you can't handle them well?
Here's a simplified example:
My company, Glance Networks, sells online visual engagement solutions (like integrated cobrowse, screen share, one-way agent video, etc.). We are a B2B business that works with large enterprises. Many of them are tech savvy. All of them do a lot of business online and want to improve customer service.
As a B2B company, our sales aren't generally completed on our website. There's a longer cycle that often extends beyond online interactions. Business is generally closed with the Sales team.
Moreover, our marketing team isn't huge, so we spend our efforts on a few channels that we can handle really well. While it's important for us to understand every area that brings interest, we focus on the ones that bring in leads that are ready for a demo.
We use a mostly Linear attribution model with some Position-based weighting.
You need to understand the following before you make a decision:
Are you selling to businesses or direct to consumers?
What is your customer's buying journey?
How many different touchpoints do you leverage?
What stage does the most heavy lifting in your business landscape?
Once you've clearly outlined this information, research the various attribution models to ensure they fit your specific business situation. Try it out and be prepared to optimize or even change it.
Top tips for marketing attribution model:
1) It needs to answer the executive question: If I spend nothing will revenue go down? Will reps or a team or the division miss their number?
2) Attribution is only good for "direction": positive, neutral, or negative ROI. It is not good for specific numbers.
3) Truly accurate attribution starts with with a "hold out" group not A/B testing or Period 1 Vs Period 2. Hold out group is like a double blind, randomized control that gets a Placebo in a science study. Meaning, this group would get no marketing or measurably different marketing than the group you are testing against. These require long periods of time with large sample sizes.
4) Attribution is good for tails. On what the top 10% of bottom 10% is working and not working. It doesn't help for optimizing.
Choosing the right attribution model will likely depend on the stage of your company and the amount of marketing channels you're pursuing. If you're company is investing in several different marketing channels and the customer journey is made up of several touch points, then it probably doesn't make sense to use a first touch or last touch attribution model. In this case, you may be better suited with a more distributed attribution model.
However, if you're a small business with a very straight forward funnel (PPC -> Landing Page -> Conversion), then you'll probably do just fine with a first click attribution method.
In my opinion, there's no right answer when it comes to choosing an attribution model. Instead, each company should evaluate the customer journey, weigh the pros and cons of each model, then choose the best one.
The best tip I can give is to define your marketing attribution model accurately. That is a complex issue: there is no one way or best way, there is no secret key. It very much depends on what is the goal, the macro goal and the micro goal. Having a clear goal or goals and being on top of your content would be a good start; the key is being very well organized.
- Invest time setting up your campaign correctly: divide the campaign into smaller compartments (the direct email part, the social media part, onsite promotion or pop-up, newsletter, etc).
- Use separate landing pages for your campaigns, with one call to action per page.
- Invest in a marketing automation tool, which will collect information about the campaign and help with lead generation and follow up. It will also help with measuring and reporting. It doesn’t have to be a full time tool, you can choose an on/off solution to keep the budget tight.- Document everything you do – analyze what works best and repeat it. When something doesn’t work, analyze what went wrong and learn the lessons from it.
Done this way there should be no problem with seeing where leads are coming from and which ways work best.
What common mistakes do marketers need to avoid?
The number one thing to avoid is the something I call “I want it all and I want it now approach”; especially for small and medium-sized businesses which have limited marketing resources.
Running multiple campaigns with different calls to action simultaneously via all possible channels definitely makes you feel really busy and gives the ‘lots is going on’ look. At the same time, this approach means you’ll have a lot of trouble measuring the campaign’s reach and effectiveness, identifying which channel brings the best result and attributing it correctly.
The second mistake comes together with the first and it is called “What a great idea, let’s do it and let’s do it now”. This is when the idea for a campaign or promotion feels so right and you are so convinced that it is one-in-the-million chance to break through that you set it all up in a rush and just hope for the best. This approach can really bite sometimes.
So my recommendation would be:
- It is better to run one or two well-planned and well-organized campaigns but to be in control of the outcomes, measure the result, tweak where needed and improve on the go.
- Every great idea has great potential to become successful. Write it down, and sleep on it. And if it feels so great in the morning, plan, share with colleagues and implement it in a calm manner.
Time is both your biggest ally and also your enemy. You absolutely MUST take time when working on campaigns and time is also a great investment. Time gives a great understanding of what is going on at every point and control over not only the campaign but also the campaign’s outcomes, which is very rewarding in the end.
I usually start with something simple like last touch attribution models. They are relatively easy to implement and most technology tools will help you get started as this is how they attribute leads and revenue. However, relatively quickly I add complexity to that model as my technological infrastructure becomes more robust and transition it to something like a multi-touch model.
Personally, I prefer the time decay model because I can map this into my MQL models which employ a similar time decay methodology. From there, I can generate a fairly accurate representation of how each channel has contributed to revenue and leads.
I don't like to use attribution models that attribute leads and revenue evenly across all channels as I believe this is misleading and deceptive when it comes time to analyze your campaign and channel performance.
In terms of avoiding mistakes, I encourage marketers to start with a model that they can implement. There is no point trying to implement a multi-touch attribution model if you don't have the necessary tools to track and report on that level of sophistication. So, start with something simple and add complexity from there.
Secondly, I always try to report channel success in the context of revenue and MQLs - this gives you greater insight into how your channels are performing at all stages of the funnel.
Finally, whenever you report on attribution results to a wider management group, always be sure to point out the advantages and disadvantages of the model you have employed. This will help them interpret the results in the report accurately.
Our company philosophy falls in line with the 'multi touch attribution model'. Every touch influences the journey however, the Dryrun team tries to isolate those 2 or 3 key points that help customers convert.
Tips for choosing a Marketing Attribution Model:
- Don't look solely at the conversion. Study and discuss the gaps between them to help you 'connect the dots' on your customers' journey.
- You can't track everyone end to end, especially in start-up phase. There is always some degree in guessing what goes on in the journey during gaps where customers appear to 'fall off the map'.
- Balance qualitative and quantitative evidence. If you rely too heavily on data without enough intuition you can put all your marketing dollars into the wrong touch leading to lots of clicks and no conversions.
Tips for Marketers:
Look at your buyers' entire journey and all their touches to identify opportunities for contact, rapport and education.
Avoid the allure of last-touch attribution models - they fall under the 'post hoc' logical fallacy - whether your customer converts has more to do with them than with your influence during a particular touch.
Know your customer and bring your intuition into the equation because clicks don't necessarily mean sales.
Today's financial SaaS tool customers have a wide market of products to try. Dryrun's entire team is focused on meeting their needs to help them feel comfortable and confident with their journey toward purchase.
Think about the buyer's journey
How does someone become aware of your brand, product or service, and then what is the process that person goes through to make a purchase decision?
This is highly important for figuring out which attribution model to use because it provides context to the different points within your funnel. Depending on how your funnel is structured, you might want to emphasize earlier or later touches more in your attribution model.
Consider your marketing channels
If you are mostly focusing on 1-2 specific channels, you should build your attribution model around those channels.
For example, if you acquire most traffic through paid ads on Facebook, which uses first-click attribution, it would make sense to use a first-click attribution model (or something similar).
Compare attribution models to learn about how different marketing channels affect your business
In Google Analytics, you can easily view your traffic data through various attribution models via Conversions --> Attribution --> Model Comparison Tool. Use this tool and the tools in the Multi-Channel Funnels tab to dive into your data and learn about how different marketing channels affect each stage of your funnel.
My top recommendation is for people to choose a marketing attribution model based on their goals.
There is no single best model to suit each and every marketer’s needs in every single situation. Of course, the last-click model is the safest, especially if you’re on a budget. But that doesn’t mean it’s best for every situation.
For example, if you have a long sales cycle, the last-click model is likely to yield minimal results. Instead, a position-based model would work much better, and help you invest in the top-performing channels.
I’d also recommend that marketers don’t hesitate to switch to a different attribution model if their current one doesn’t seem to suit them. I think this is where many marketers make a mistake.
When considering a change, they might be held back by the fear of failing again with the next model they choose. So instead, they decide to stick with what’s familiar instead of testing out other models to see which one works best.
Try analyzing your customers’ behavior to see how your current attribution model is working, and how many touchpoints they go through before converting. You can then use this data to determine which model will be most right for you.
Attribution for app installs in particular is a particularly tough game. Unfortunately the standard tracking parameters we include on web URLs such as utm_source don’t survive the app store. This means when the app is opened for the first time, we’re unable to send through any of that tracking data to an attribution server.
The good news is that Deep linking technologies such as Branch help to solve this. They offer dynamic links that store parameters that survive the app store.
The last click attribution model is my favourite. The reason for this is that I believe the advert or message that drives that final point of conversion is ultimately the one that should take most of the kudos. Additionally, attributing some of the conversion to low impact mediums such as display feels a little irrelevant (we’re going to run re-marketing anyways).
Common mistakes to look for include:
- incorrectly referencing the same source in your url tracking parameters
- blocking url tracking parameters in Google Web Master tools
- content grouping of dissimilar sources/mediums
Top attribution model selection tips:
1) Figure out what you want to solve for in terms of optimizing your campaign mix. Are you looking for campaigns that add new leads to your database? First touch model.
Are you looking for campaigns that qualify leads into a real $ opportunity down your funnel? Last touch (marketing qualified) model.
Are you a consumer business looking for campaigns responsible for the last click or shopping cart checkout? Last click / last touch (pre-checkout) model.
Trying to split attribution across all campaigns that engaged with leads? Use even-spread multi-touch (beware that even-spread attribution can be misinterpreted, see #3).
2) Run multiple models and compare the differing results, especially since different models tell you different things. (See my resources below for a summary of the differences)
3) Be more investigative about the attributed revenue and what the attribution actually means.
Example: In an even-spread model, campaigns that are most frequently used and touch the highest number of leads could be overly credited with revenue attribution. Even-spread is not adjusted for campaigns that are unequal in revenue impact and importance.
Being more investigative allows you to dive into the campaigns and truly assess whether you want to give a campaign that much credit. Then, you can optimize according to the conclusions from the "venn diagram" of attribution models you use.
See my resources below for references on attribution models and methodologies for selecting them.
For more help or advice, feel free to email me at my first name dot last name at fullcircleinsights dot com.
Know which models fit your business, don't just grab one that looks good.
Brief your sales team on the attribution management - that will help them understand the customer journey when they place the first call.
Remember - Lead Source is a Lie!
Here are my tips:
1- Create a customer journey map to brainstorm touchpoints and channels they will use at different stages of the buying cycle.
Choose an attribution model you are confident in the data and you believe in.
If you confident in the data, you’ll be confident in the attribution model and you’ll be happy to scale and scale quickly.
Include assisted conversion path in your attribution model. If we only ever ran marketing campaigns that provided an ROI for 1 day direct click, we’d never spend any money on marketing. A conversion may take a brand campaign, a direct click and remarketing, so it’s important to understand the role of each.
What common mistakes do marketers need to avoid?
A common mistake is choosing one attribution model, sticking with it and being 100% committed to the model. Different models will tell weight importance on different metrics, which may provide you a better insight for certain lead sources.
When thinking about what is the right marketing attribution model, it really depends on your business.
In enterprise software sales, it's a complex research and buying process that typically takes multiple activities and channels to attain both breadth across numerous individuals at the same target company as well as depth of providing valuable content to those prospects to move them from the research to buying stage.
While there of course is the purchase decision-maker, there are usually other influencers and recommenders that should be accounted for in your marketing outreach and in terms of your attribution.
In terms of the steps of the lead to sale cycle, first touch, lead conversion and opportunity creation are important aspects to track, but tying 100% of your attribution to any of these stages by themselves oversimplifies the marketing effort in what is typically a longer sales cycle.
A more comprehensive attribution model looks at the several key points within the entire cycle and incorporates the different contacts at the target company with whom you have touched. It starts with awareness/that first touch that you can track, then what helped drive to lead conversion and then to opportunity creation.
Depending on the setup of your sales and marketing teams you could add the final step of customer close, but when talking about new sales (as opposed to up-selling/cross-selling existing customers) that customer closing step is usually more in the hands of sales and a delivery/technical team.
Treating all campaigns and businesses in the same way to determine value attribution, is a very common mistake that marketers need to avoid. In selecting the right marketing attribution model it's critical to consider the type of campaign you are running and it’s user journey.
For very short one touch journeys, like lead generation via paid search direct to a landing page, last click attribution makes good sense.
For multi-touch journeys, (such as high value purchases that require user research or exposure over time) a position based attribution is more appropriate as it gives more weight to first click and last click interactions.
The first interaction is what created interest in the user to begin with and the last is what motivated the user to convert or take action.
The biggest mistake most marketers make when choosing a marketing attribution model is they ignore some of the principles of marketing.
The principles of marketing have not changed. What is forever evolving is technology which means marketers are now spoiled for choice due to the rapid growth in online marketing & advertising channels.
Direct response marketing is the queen of marketing because it’s based on the ability to track and measure results. And now companies have access to significantly more data to track and measure the effectiveness of branding and marketing campaigns, it’s even more powerful.
The trap is when marketers fail to identify the right metrics for success. You must measure metrics you can control to enable continual adjustment to improve results - otherwise there's no point. ROI remains the arbiter of success.
The type of attribution model will depend on two primary factors:
1. Customer journey touchpoints
2. Campaign objectives
While attribution modeling can be a daunting topic for many marketers and business owners, you can simplify it by first thinking about the number of touchpoints along the customer journey.
If you're running a basic direct response campaign that uses ads to drive traffic to a landing page to convert users, a simple last-touch model works. In general, where there is little buying consideration, a last-touch attribution model works as you don't need to assign "credit" to first and middle touch channels.
Similary, if you're primary campaign objective is brand awareness, a first touch attribution model will work. This way you are assigning more conversion value to initial interactions that kicked off the conversion pathway.
However, if you're running a campaign that has multiple touchpoints (Eg. Social Media --> AdWords ---> Organic --> Direct ) you'll need to look at using a more weitghed attribution model.
You can use a linear model to assign equal value to each touchpoint, use a time decay model (good for timed promotional campaigns), or a more customizable position-based model to assign specific value to every touchpoint. This article does a good job at looking at the pros and cons of each.
Always approach attribution modeling by first thinking about the number of campaign touchpoints and overarching campaign objectives. Once you've answered those two questions the appropriate model will be clear.
Buyer interest is influenced by total marketing impressions - which includes multiple touch points that happen both online and offline.
While online efforts are generally easier to track than offline, understanding the role of inside sales and the optimum time for follow-up calls as well as the effectiveness of those outbound calls is much more difficult. That's why one of the hardest parts about attribution models is syncing online efforts with offline efforts.
Where possible, use a customizable attribution model that pulls CRM call status information into the total attribution equation and includes offline touches as a core input (such as outbound or inbound sales calls) into the model. This information can be extremely valuable in understanding the relative importance of each touchpoint and helps you fine-tune the roll of inside sales into the total buyer journey.
Even with the best intentions, no attribution model is perfect.
Multi-touch models include user bias and opinion that will seldom past muster of a committee. While no model survives contact with a committee, any multi-touch model is still far better than a simple last touch attribution model.
First and foremost - track for ROI and not for anything else, meaning impressions, email opens, clicks etc are important for bottleneck optimizations (if clicks are high but conversions are low - you know there is a mismatch between the AD and the LANDING PAGE text, or the landing page just sucks)
In b2b - we track all the way to the opportunity level and then we run multi-variant testing on the entire matrix to determine what marketing experiments yield a positive ROI (usually 15-20% of the total mix), we eliminate negative ROI, reduce resources for breakeven and pile up on strong, positive ROI experiments.
I highly recommend using hidden form fields and pass to those fields all the possible UTM tags you can possibly get, including the campaign name, campaign channel, campaign landing page variation, campaign creative, campaign audience and anything else that might change results for you
Finally - I propose getting the final and most important feedback point from sales - for every lead generated you need to know the quality of that lead.
Lead scoring can help, but manually sitting down with the head of sales/ae and scoring 100 leads at a time can really change your model and give you more insight into your marketing programs than anything else.
Of course the ultimate score is a salesforce opportunity created. show a full funnel analysis for the CMO (attached from Metadata):
Choosing a marketing attribution model for your company depends a lot on your line of business, and what you'll actually do with the data you get. It's easy to say "we want to be more data-driven" and then capture a bunch of data, but how you parse that data is the hard part!
Here, I think backing out to the high-level goals of your company, team and strategy is really important. Figure out what the most useful data is for helping you achieve those goals.
Does looking at last-click attribution help you achieve your short and long term goals? Is that what you should be looking at every morning and basing your budget decisions on? If you're focused on that, how are you also thinking about other goals, other channels and the long-term impact of PR, brand development, awareness, and inbound strategies?
For entrepreneurs and small companies, I think starting simply is much better than having no data at all, and giving yourself permission to start simply can help you from getting bogged down in the decision making and process of implementing a larger model and tracking system.
The right attribution model is going to be unique for each business based on their buying cycle and advertising strategy.
To find the best attribution model for your multi-channel marketing, I can only offer one major piece of advice. You must must must read carefully through Avinash Kaushik's stellar article on the topic and following his instructions based on your buying cycle. His actionable method has been a great resource and many would benefit from it.
There you have it - the key tips (and mistakes) you need to know when choosing the right marketing attribution model for your business.
Remember what analytics thought leader Avinash Kaushik stated:
…every attribution model has built into it biases and opinions that often struggle to stand any intellectual scrutiny, or the simple laws of common sense.
Every attribution model has it's pros and cons. The most important thing you can do is test, predict and measure results on a regular basis to reach your goal: understand the true impact on each channel and campaign across customers lifecycle.